Photovoltaic power station power generation power prediction method, device and electronic equipment

Abstract

The application provides a photovoltaic power station power generation power prediction method, device and electronic equipment, and belongs to the technical field of photovoltaic power stations. The method comprises the following steps: acquiring operation state data and weather forecast data of a plurality of photovoltaic power stations; constructing a graph data structure based on the operation state data and the weather forecast data of the plurality of photovoltaic power stations; inputting the graph data structure into a power generation power prediction model to obtain a power generation power prediction result of a target photovoltaic power station in the plurality of photovoltaic power stations output by the power generation power prediction model; wherein the power generation power prediction model is constructed based on a graph convolutional neural network; and the power generation power prediction model is obtained by training based on sample operation state data, sample weather forecast data of a plurality of associated photovoltaic power stations and a power generation power label of the target photovoltaic power station in the plurality of associated photovoltaic power stations. The application is used to solve the defects of the prior art, i.e., the limited capturing ability of numerical weather forecast errors and the incomprehensive consideration of data.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power plant technology, specifically to a method for predicting the power generation of a photovoltaic power plant, a device for predicting the power generation of a photovoltaic power plant, an electronic device, a machine-readable storage medium, and a computer program product. Background Technology

[0002] With the increasing global emphasis on renewable energy, photovoltaic power generation, as a clean and renewable energy form, is becoming increasingly important for the stable operation, optimized dispatch, and energy management of power systems, especially as the installed capacity and grid connection ratio of photovoltaic power generation increase.

[0003] Currently, photovoltaic (PV) power prediction technology typically uses numerical weather prediction data of the PV power plant's latitude and longitude as input to establish a mapping model from the numerical weather prediction data to power generation, thereby predicting the future power generation of the PV power plant. The algorithms used in this mapping model usually include neural network algorithms, support vector machine algorithms, and linear regression algorithms. Finally, based on the established mapping model, calculations are performed using the numerical weather prediction data for the time to be predicted to obtain the power prediction result for the PV power plant at that time.

[0004] Existing technologies suffer from the following problems: the mapping model input does not consider cloud information in the surrounding area and the data considered is not comprehensive enough. Because numerical weather prediction inherently involves forecast errors, the power mapping model needs to fit the characteristics of these errors to a certain extent to reduce their impact on the accuracy of power generation prediction. Similarly, when the mapping model considers insufficient data, it also leads to a decrease in the accuracy of power generation prediction. Therefore, existing technologies have limited ability to capture numerical weather prediction errors and suffer from insufficient data consideration. Summary of the Invention

[0005] The purpose of this invention is to provide a method, apparatus, and electronic device for predicting the power generation of a photovoltaic power plant, in order to address the shortcomings of existing technologies, such as limited ability to capture numerical weather forecast errors and insufficient data coverage.

[0006] To achieve the above objectives, embodiments of the present invention provide a method for predicting the power generation of a photovoltaic power plant, comprising: Acquire operational status data and weather forecast data from multiple photovoltaic power plants; A graph data structure is constructed based on the operational status data and weather forecast data of the multiple photovoltaic power plants; The graph data structure is input into the power generation prediction model to obtain the power generation prediction result of one target photovoltaic power station among multiple photovoltaic power stations output by the power generation prediction model. The power generation prediction model is constructed based on a graph convolutional neural network. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. The multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0007] Optionally, the power generation prediction model is trained through the following steps: Acquire sample operational status data and sample weather forecast data from multiple initial photovoltaic power plants; The correlation coefficient between sample weather forecast data of multiple initial photovoltaic power plants is calculated based on the Pearson correlation analysis method or the Spearman correlation method, and multiple associated photovoltaic power plants with a correlation coefficient greater than or equal to a set threshold are identified from the multiple correlation coefficients. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants.

[0008] Optionally, the construction of the graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants includes: Nodes are constructed based on each of the multiple photovoltaic power plants; The node characteristics of each node are constructed based on the operating status data and weather forecast data of each photovoltaic power station; An adjacency matrix is ​​constructed by establishing connection weights between different nodes in the multiple photovoltaic power plants; the connection weights represent the connection relationships and the degree of mutual correlation between the nodes.

[0009] Optionally, in the power generation prediction model, the mapping relationship between input data and output data is expressed by the following formula: ; Where Y represents the output data, X represents the input data, ReLU() represents the linear rectified function, and softmax() represents the normalized exponential function. W （0） This represents the mapping weights from the input layer to the hidden layer in a graph convolutional neural network structure. W （1） This represents the mapping weights from hidden layers to the output layer in a graph convolutional neural network structure. This represents the adjacency matrix.

[0010] Optionally, the operating status data includes at least one of the following: device temperature, device current, and device voltage.

[0011] Optionally, the weather forecast data includes at least one of latitude and longitude, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, and pressure at different heights.

[0012] On the other hand, embodiments of the present invention also provide a photovoltaic power plant power generation prediction device, comprising: The acquisition module is used to acquire operational status data and weather forecast data from multiple photovoltaic power plants. The construction module is used to construct a graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants; The prediction module is used to input the graph data structure into the power generation prediction model to obtain the power generation prediction result of one target photovoltaic power station among multiple photovoltaic power stations output by the power generation prediction model. The power generation prediction model is constructed based on a graph convolutional neural network. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. The multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0013] Optionally, the power generation prediction model is trained through the following steps: Acquire sample operational status data and sample weather forecast data from multiple initial photovoltaic power plants; The correlation coefficient between sample weather forecast data of multiple initial photovoltaic power plants is calculated based on the Pearson correlation analysis method or the Spearman correlation method, and multiple associated photovoltaic power plants with a correlation coefficient greater than or equal to a set threshold are identified from the multiple correlation coefficients. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants.

[0014] Optionally, the construction of the graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants includes: Nodes are constructed based on each of the multiple photovoltaic power plants; The node characteristics of each node are constructed based on the operating status data and weather forecast data of each photovoltaic power station; An adjacency matrix is ​​constructed by establishing connection weights between different nodes in the multiple photovoltaic power plants; the connection weights represent the connection relationships and the degree of mutual correlation between the nodes.

[0015] Optionally, in the power generation prediction model, the mapping relationship between input data and output data is expressed by the following formula: ; Where Y represents the output data, X represents the input data, ReLU() represents the linear rectified function, and softmax() represents the normalized exponential function. W （0） This represents the mapping weights from the input layer to the hidden layer in a graph convolutional neural network structure. W （1） This represents the mapping weights from hidden layers to the output layer in a graph convolutional neural network structure. This represents the adjacency matrix.

[0016] On the other hand, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described photovoltaic power generation prediction method.

[0017] On the other hand, the present invention also provides a machine-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described photovoltaic power generation prediction method.

[0018] On the other hand, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-mentioned photovoltaic power generation prediction method.

[0019] Through the above technical solution, the power generation prediction model of this invention is constructed based on a graph convolutional neural network. The model is trained using sample operational status data from multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among these associated plants. Therefore, the graph convolutional neural network of this invention integrates operational status data and weather forecast data from multiple photovoltaic power plants for training. This allows for the extraction of cloud information interaction relationships and cloud movement characteristics at the latitude and longitude of the photovoltaic power plants, fusing spatial relationships with the meteorological parameter characteristics of each plant's latitude and longitude, thereby improving the model's expressive and predictive capabilities. By mining the mutual characteristics of weather forecast data from multiple photovoltaic power plants, the ability to capture numerical weather forecast errors is improved, the data considered is more comprehensive, and the accuracy of power prediction is enhanced.

[0020] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart illustrating the photovoltaic power generation prediction method provided by the present invention. Figure 2 This is a schematic diagram of the graph data structure provided by the present invention; Figure 3 This is one of the structural schematic diagrams of the power generation prediction model provided by the present invention; Figure 4 This is the second schematic diagram of the power generation prediction model provided by the present invention; Figure 5 This is a schematic diagram of the photovoltaic power generation prediction device provided by the present invention; Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0022] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0023] Current day-ahead power prediction technologies for photovoltaic (PV) power plants typically use numerical weather prediction (NMR) data of the PV power plant's latitude and longitude as input to establish a mapping model from the NMR data to power generation, predicting the future power output of the power plant. The key steps in day-ahead power prediction for PV power plants include: First, acquiring NMR data. NMR data is the prediction of future meteorological parameters based on meteorological evolution principles, including latitude and longitude, irradiance, cloud cover, wind speed, wind direction, temperature, humidity, and pressure at various altitudes. NMR data can be obtained by purchasing relevant products or by building a NMR system for calculation. Second, processing the NMR data, including selecting optimal parameters and parameter transformation, to obtain the mapping model input most relevant to power generation. Third, using specific mathematical algorithms to establish the mapping model from the model input to the power generation; the algorithms typically used include neural network algorithms, support vector machine algorithms, and linear regression algorithms. Finally, based on the established mapping model, calculations are performed using the NMR data for the time to be predicted to obtain the power prediction result for the PV power plant at that time.

[0024] The above methods have the following problems: the mapping model input does not consider cloud information in the surrounding area and the data considered is not comprehensive enough. Because numerical weather prediction has prediction errors, the power mapping model needs to fit the characteristics of these errors to a certain extent to reduce their impact on the accuracy of power generation prediction. Similarly, when the mapping model considers insufficient data, it also leads to a decrease in the accuracy of power generation prediction. Therefore, existing technologies have limited ability to capture numerical weather prediction errors and insufficient data consideration.

[0025] Therefore, the purpose of this invention is to provide a method, apparatus, and electronic device for predicting the power generation of a photovoltaic power plant, in order to solve the shortcomings of the prior art, such as limited ability to capture numerical weather forecast errors and insufficient consideration of data.

[0026] Method Implementation Examples Please refer to Figure 1 This invention provides a method for predicting the power generation of a photovoltaic power plant, comprising: Step 100: Obtain operational status data and weather forecast data from multiple photovoltaic power plants.

[0027] Electronic equipment acquires operational status data and weather forecast data from multiple photovoltaic power plants. The operational status data includes at least one of the following: equipment temperature, equipment current, and equipment voltage. The equipment can be inverters, transformers, or other devices within the photovoltaic power plant. The weather forecast data includes at least one of the following: the latitude and longitude of the photovoltaic power plant, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, and pressure at different altitudes.

[0028] It should be noted that the embodiments of the present invention are aimed at the day-ahead power generation forecast of photovoltaic power plants. The grid dispatching agency formulates various power output plans based on the day-ahead power generation forecast. The accuracy of the day-ahead power generation forecast has a very important impact on the stable operation of the grid.

[0029] Step 200: Construct a graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power stations.

[0030] The electronic device constructs a graph data structure based on the operational status data and weather forecast data of the multiple photovoltaic power plants. This graph data structure is used as the input to a graph convolutional neural network.

[0031] In one embodiment, constructing a graph data structure based on the operating status data and weather forecast data of the plurality of photovoltaic power plants includes: constructing a node based on each of the plurality of photovoltaic power plants; constructing node features for each node based on the operating status data and weather forecast data of each photovoltaic power plant; constructing an adjacency matrix by constructing connection weights between different nodes in the plurality of photovoltaic power plants; wherein the connection weights represent the connection relationships and the degree of mutual correlation between the nodes.

[0032] This invention can construct the association relationship between multiple photovoltaic power plants based on a graph data structure. The process of constructing the graph data structure includes node definition, node feature definition, adjacency matrix construction, and graph structure modeling.

[0033] Node definition refers to using photovoltaic (PV) power stations as graph nodes, with each node representing a PV power station. Node features refer to the operational status data of the PV power station and the weather forecast data for its latitude and longitude as the characteristics of the graph nodes. Node features include equipment temperature, equipment current, equipment voltage, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, and pressure at various altitudes. The adjacency matrix refers to the relationships between PV power stations, reflecting their geographical location, grid connection, and other relationships. Graph structure modeling refers to designing the graph data structure for multiple PV power station nodes; please refer to [reference needed]. Figure 2 The illustrated embodiment of the present invention presents a graph data structure for multiple photovoltaic power station nodes (taking six photovoltaic power stations as an example). Each node represents a photovoltaic power station, and each node's attributes include selected parameters, including meteorological and measurement parameters, such as: equipment temperature, equipment current, equipment voltage, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, pressure, and other numerical weather forecast parameters at different heights, as well as historical measurement data from wind towers at different heights. Different nodes are interconnected through connection weights, which represent the connection relationship and degree of interrelation between nodes. This graph data structure can reflect the local meteorological characteristics of the stations and the spatial correlation characteristics between different photovoltaic power stations, exhibiting strong expressive power.

[0034] Step 300: Input the graph data structure into the power generation prediction model to obtain the power generation prediction result of one target photovoltaic power station among the multiple photovoltaic power stations output by the power generation prediction model; The power generation prediction model is constructed based on a graph convolutional neural network. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. The multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0035] Sample operational status data and sample weather forecast data can be historical operational status data and historical weather forecast data of photovoltaic power plants. The power generation label is the actual power generation value of the photovoltaic power plant. Multiple associated photovoltaic power plants refer to multiple photovoltaic power plants whose weather forecast data are correlated within a geographical area. Multiple initial photovoltaic power plants are selected through correlation analysis to identify multiple associated photovoltaic power plants with high weather correlation. The graph convolutional neural network in this embodiment of the invention, based on the graph data structure of step 200, uses graph convolution technology to establish a mapping relationship between graph data, thereby predicting the power generation curve of a specific power plant. Please refer to... Figure 3 The model structure of the convolutional neural network when the prediction result is for a single power station is shown in the figure. Figure 3 X1 to X6 represent the operational status data and weather forecast data for the six photovoltaic power plants, while Y1 represents the predicted power generation of a single photovoltaic power plant among the six. It should be noted that please refer to... Figure 4 In other embodiments, the power generation prediction model can be generated by inputting the operating status data and weather forecast data of multiple photovoltaic power plants into the power generation prediction model, and outputting the power generation prediction results of multiple photovoltaic power plants at the same time. Figure 4 X1 to X6 represent the operating status data and weather forecast data of the six photovoltaic power plants, while Y1 to Y6 represent the power generation prediction results of the six photovoltaic power plants.

[0036] The graph convolutional model structure in this embodiment of the invention includes an input layer, a hidden layer, and an output layer. Unlike traditional MLPs (Multilayer Perceptrons), the weights between different layers include not only mapping weights but also connection weights between nodes, represented by an adjacency matrix or a generalized adjacency matrix. In the power generation prediction model of this embodiment, the mapping relationship between input data and output data is expressed by the following formula: ; Where Y represents the output data, X represents the input data, ReLU() represents the linear rectified function, and softmax() represents the normalized exponential function. W （0） This represents the mapping weights from the input layer to the hidden layer in a graph convolutional neural network structure. W （1） This represents the mapping weights from hidden layers to the output layer in a graph convolutional neural network structure. This represents the adjacency matrix.

[0037] When forecasting power generation, weather forecast data and operational status data from multiple photovoltaic power plants at the time of forecast are organized into a format such as... Figure 2The data is formatted as input to a graph convolutional neural network structure to calculate the predicted power generation of a single photovoltaic power station at the corresponding time. Based on the node graph data structure constructed in this embodiment, the correlation between spatial irradiance and cloud cover can be established on the basis of characterizing the meteorological characteristics of a single power station, thereby improving the data's expressive power.

[0038] The power generation prediction model of this invention is constructed based on a graph convolutional neural network. This model is trained using sample operational status data from multiple associated photovoltaic (PV) power plants, sample weather forecast data, and the power generation labels of the target PV power plant among these associated plants. Therefore, the graph convolutional neural network of this invention, trained by integrating operational status data and weather forecast data from multiple PV power plants, can extract the cloud information interaction relationships and cloud movement characteristics of the latitude and longitude of the PV power plants, fusing spatial relationships with the meteorological parameter characteristics of each plant's latitude and longitude, thereby improving the model's expressive and predictive capabilities. By mining the mutual characteristics of weather forecast data from multiple PV power plants, the ability to capture numerical weather forecast errors is improved, the data considered is more comprehensive, and the accuracy of power prediction is enhanced.

[0039] In other aspects of the embodiments of the present invention, the power generation prediction model is trained through the following steps: Step 10: Obtain sample operating status data and sample weather forecast data for multiple initial photovoltaic power plants; Step 20: Calculate the correlation coefficient between sample weather forecast data of multiple initial photovoltaic power plants based on the Pearson correlation analysis method or the Spearman correlation method, and determine multiple associated photovoltaic power plants whose correlation coefficient is greater than or equal to a set threshold from the multiple correlation coefficients; Step 30: Train the power generation prediction model based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants.

[0040] Electronic devices acquire historical operational status data and historical weather forecast data from multiple initial photovoltaic power plants in the same geographical area as sample operational status data and sample weather forecast data. Then, they calculate the correlation coefficient between the sample weather forecast data from the multiple initial photovoltaic power plants based on Pearson correlation analysis or Spearman correlation analysis. Taking Pearson correlation analysis as an example, the formula for calculating the Pearson correlation coefficient r is as follows: ; in, n It refers to the number of samples. x i and y i These are two variables at the thi The values ​​taken from each sample. and They are x and y The sample mean is calculated using the following formulas: and .For example x i ... x n The characteristics of each node of photovoltaic power station A are as follows (equipment temperature, equipment current, equipment voltage, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, pressure, etc. at each altitude). express x i ... x n The average of all data, y i ... y n The characteristics of each node of photovoltaic power station B are as follows (equipment temperature, equipment current, equipment voltage, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, pressure, etc. at each altitude). Indicates y i ... y n The average value of all data is used. The correlation coefficient between photovoltaic power station A and photovoltaic power station B can be obtained by calculating the above formula. If the correlation coefficient is greater than or equal to 0.9, photovoltaic power station A and photovoltaic power station B are determined to be related photovoltaic power stations. Therefore, by calculating the correlation of other initial photovoltaic power stations from multiple initial photovoltaic power stations using the above Pearson correlation analysis method, multiple related photovoltaic power stations can be screened out from the multiple initial photovoltaic power stations.

[0041] Taking the Spearman correlation method as an example, the Spearman correlation coefficient ρ The calculation formula is as follows: ; in, ρ This represents the Spearman correlation coefficient. d i It is the first i The absolute value of the difference between the ranks of two variables in an observation. n This represents the total number of observations. The calculation steps typically include: sorting the nodal characteristics of photovoltaic power plant A and photovoltaic power plant B as observations, and assigning them ranks. The rank is the position of the observation in the sorted list. If multiple observations are equal, their rank is the ranking of the average of these observations. The absolute value d of the difference between the ranks of each pair of observations is then calculated. i Calculate all d i 2 The sum of, i.e. The Spearman correlation coefficient ρ is calculated using the formula. The correlation coefficient between photovoltaic power station A and photovoltaic power station B can be obtained by calculating the above formula. If the correlation coefficient is greater than or equal to 0.9, photovoltaic power station A and photovoltaic power station B are determined to be related photovoltaic power stations. Therefore, by calculating the correlation of other initial photovoltaic power stations from multiple initial photovoltaic power stations using the above Spearman correlation analysis method, multiple related photovoltaic power stations can be screened out from the multiple initial photovoltaic power stations.

[0042] This invention further trains the power generation prediction model based on sample operational status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. During training, a graph data structure constructed from sample operational status data and sample weather forecast data of multiple associated photovoltaic power plants is input into a graph convolutional neural network to obtain prediction results. Based on the prediction results and the true labels, a loss function value is calculated to evaluate the model performance. Then, based on the loss function value, the gradient of the network parameters is calculated using optimization algorithms such as gradient descent, and the network weights are updated. This process of calculating loss and backpropagation is repeated until a preset number of iterations is reached or the loss function value converges. This results in a well-trained power generation prediction model.

[0043] Since multiple associated photovoltaic power stations are those among multiple initial photovoltaic power stations whose weather correlation exceeds a set threshold, training a graph convolutional neural network (GNN) based on sample operational status data and sample weather forecast data from these associated power stations is beneficial for further improving the data quality input to the GNN. This facilitates modeling based on graph data using graph convolution techniques, integrates the correlations between power stations, extracts the relationships between meteorological data at different spatial locations, improves the model's ability to express and capture numerical weather forecast errors, enhances the model's mapping capabilities, and improves prediction accuracy.

[0044] Device Examples Please refer to Figure 5 On the other hand, embodiments of the present invention also provide a photovoltaic power plant power generation prediction device, comprising: The acquisition module 501 is used to acquire operating status data and weather forecast data from multiple photovoltaic power plants; Module 502 is used to construct a graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants; Prediction module 503 is used to input the graph data structure into the power generation prediction model to obtain the power generation prediction result of one target photovoltaic power station among multiple photovoltaic power stations output by the power generation prediction model; The power generation prediction model is constructed based on a graph convolutional neural network. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. The multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0045] Optionally, the power generation prediction model is trained through the following steps: Acquire sample operational status data and sample weather forecast data from multiple initial photovoltaic power plants; The correlation coefficient between sample weather forecast data of multiple initial photovoltaic power plants is calculated based on the Pearson correlation analysis method or the Spearman correlation method, and multiple associated photovoltaic power plants with a correlation coefficient greater than or equal to a set threshold are determined from the multiple correlation coefficients. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants.

[0046] Optionally, the construction of the graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants includes: Nodes are constructed based on each of the multiple photovoltaic power plants; The node characteristics of each node are constructed based on the operating status data and weather forecast data of each photovoltaic power station; An adjacency matrix is ​​constructed by establishing connection weights between different nodes in the multiple photovoltaic power plants; the connection weights represent the connection relationships and the degree of mutual correlation between the nodes.

[0047] Optionally, in the power generation prediction model, the mapping relationship between input data and output data is expressed by the following formula: ; Where Y represents the output data, X represents the input data, ReLU() represents the linear rectified function, and softmax() represents the normalized exponential function. W （0） This represents the mapping weights from the input layer to the hidden layer in a graph convolutional neural network structure. W （1） This represents the mapping weights from hidden layers to the output layer in a graph convolutional neural network structure. This represents the adjacency matrix.

[0048] Optionally, the operating status data includes at least one of the following: device temperature, device current, and device voltage.

[0049] Optionally, the weather forecast data includes at least one of latitude and longitude, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, and pressure at different heights.

[0050] The photovoltaic power generation prediction device includes a processor and a memory. The acquisition module 501, the construction module 502, and the prediction module 503 are all stored in the memory as program units. The processor executes the program units stored in the memory to realize the corresponding functions.

[0051] A processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured.

[0052] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0053] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include: a processor 610, a communication interface 620, a memory 630, and a communication bus 640, wherein the processor 610, the communication interface 620, and the memory 630 communicate with each other through the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a photovoltaic power generation prediction method. This method includes: acquiring operating status data and weather forecast data of multiple photovoltaic power plants; constructing a graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants; inputting the graph data structure into a power generation prediction model to obtain the power generation prediction result of a target photovoltaic power plant among the multiple photovoltaic power plants output by the power generation prediction model; wherein the power generation prediction model is constructed based on a graph convolutional neural network; the power generation prediction model is trained based on sample operating status data and sample weather forecast data of multiple associated photovoltaic power plants, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants; the multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0054] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0055] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a machine-readable storage medium. When the computer program is executed by a processor, the computer can execute a photovoltaic power generation prediction method. The method includes: acquiring operating status data and weather forecast data of multiple photovoltaic power plants; constructing a graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants; inputting the graph data structure into a power generation prediction model to obtain a power generation prediction result of a target photovoltaic power plant among the multiple photovoltaic power plants output by the power generation prediction model; wherein, the power generation prediction model is constructed based on a graph convolutional neural network; the power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants; the multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0056] In another aspect, the present invention also provides a machine-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for predicting the power generation of a photovoltaic power plant. This method includes: acquiring operational status data and weather forecast data of multiple photovoltaic power plants; constructing a graph data structure based on the operational status data and weather forecast data of the multiple photovoltaic power plants; inputting the graph data structure into a power generation prediction model to obtain a power generation prediction result for a target photovoltaic power plant among the multiple photovoltaic power plants output by the power generation prediction model; wherein the power generation prediction model is constructed based on a graph convolutional neural network; the power generation prediction model is trained based on sample operational status data and sample weather forecast data of multiple associated photovoltaic power plants, as well as the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants; the multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

[0057] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0058] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the power generation of a photovoltaic power plant, characterized in that, include: Acquire operational status data and weather forecast data from multiple photovoltaic power plants; A graph data structure is constructed based on the operational status data and weather forecast data of the multiple photovoltaic power plants; The graph data structure is input into the power generation prediction model to obtain the power generation prediction result of one target photovoltaic power station among multiple photovoltaic power stations output by the power generation prediction model. The power generation prediction model is constructed based on a graph convolutional neural network. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. The multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

2. The photovoltaic power generation prediction method according to claim 1, characterized in that, The power generation prediction model is trained through the following steps: Acquire sample operational status data and sample weather forecast data from multiple initial photovoltaic power plants; The correlation coefficient between sample weather forecast data of multiple initial photovoltaic power plants is calculated based on the Pearson correlation analysis method or the Spearman correlation method, and multiple associated photovoltaic power plants with a correlation coefficient greater than or equal to a set threshold are determined from the multiple correlation coefficients. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants.

3. The photovoltaic power generation prediction method according to claim 1, characterized in that, The graph data structure constructed based on the operational status data and weather forecast data of the multiple photovoltaic power plants includes: Nodes are constructed based on each of the multiple photovoltaic power plants; The node characteristics of each node are constructed based on the operating status data and weather forecast data of each photovoltaic power station; An adjacency matrix is ​​constructed by establishing connection weights between different nodes in the multiple photovoltaic power plants; the connection weights represent the connection relationships and the degree of mutual correlation between the nodes.

4. The photovoltaic power generation prediction method according to claim 3, characterized in that, In the power generation prediction model, the mapping relationship between input data and output data is expressed by the following formula: ； Where Y represents the output data, X represents the input data, ReLU() represents the linear rectified function, and softmax() represents the normalized exponential function. W （0） This represents the mapping weights from the input layer to the hidden layer in a graph convolutional neural network structure. W （1） This represents the mapping weights from hidden layers to the output layer in a graph convolutional neural network structure. This represents the adjacency matrix.

5. The photovoltaic power generation prediction method according to any one of claims 1 to 4, characterized in that, The operating status data includes at least one of the following: equipment temperature, equipment current, and equipment voltage.

6. The photovoltaic power generation prediction method according to any one of claims 1 to 4, characterized in that, The weather forecast data includes at least one of the following: latitude and longitude, irradiance, cloud cover rate, wind speed, wind direction, temperature, humidity, and pressure at different heights.

7. A photovoltaic power plant power generation prediction device, characterized in that, include: The acquisition module is used to acquire operational status data and weather forecast data from multiple photovoltaic power plants. The construction module is used to construct a graph data structure based on the operating status data and weather forecast data of the multiple photovoltaic power plants; The prediction module is used to input the graph data structure into the power generation prediction model to obtain the power generation prediction result of one target photovoltaic power station among multiple photovoltaic power stations output by the power generation prediction model; The power generation prediction model is constructed based on a graph convolutional neural network. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation labels of the target photovoltaic power plant among the multiple associated photovoltaic power plants. The multiple associated photovoltaic power plants are obtained by filtering sample weather forecast data of multiple initial photovoltaic power plants through correlation analysis.

8. The photovoltaic power generation prediction device according to claim 7, characterized in that, The power generation prediction model is trained through the following steps: Acquire sample operational status data and sample weather forecast data from multiple initial photovoltaic power plants; The correlation coefficient between sample weather forecast data of multiple initial photovoltaic power plants is calculated based on the Pearson correlation analysis method or the Spearman correlation method, and multiple associated photovoltaic power plants with a correlation coefficient greater than or equal to a set threshold are determined from the multiple correlation coefficients. The power generation prediction model is trained based on sample operating status data of multiple associated photovoltaic power plants, sample weather forecast data, and the power generation label of the target photovoltaic power plant among the multiple associated photovoltaic power plants.

9. The photovoltaic power generation prediction device according to claim 7, characterized in that, The graph data structure constructed based on the operational status data and weather forecast data of the multiple photovoltaic power plants includes: Nodes are constructed based on each of the multiple photovoltaic power plants; The node characteristics of each node are constructed based on the operating status data and weather forecast data of each photovoltaic power station; An adjacency matrix is ​​constructed by establishing connection weights between different nodes in the multiple photovoltaic power plants; the connection weights represent the connection relationships and the degree of mutual correlation between the nodes.

10. The photovoltaic power generation prediction device according to claim 9, characterized in that, In the power generation prediction model, the mapping relationship between input data and output data is expressed by the following formula: ； Where Y represents the output data, X represents the input data, ReLU() represents the linear rectified function, and softmax() represents the normalized exponential function. W （0） This represents the mapping weights from the input layer to the hidden layer in a graph convolutional neural network structure. W （1） This represents the mapping weights from hidden layers to the output layer in a graph convolutional neural network structure. This represents the adjacency matrix.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the photovoltaic power generation prediction method according to any one of claims 1 to 6.

12. A machine-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the photovoltaic power generation prediction method according to any one of claims 1 to 6.

13. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the photovoltaic power generation prediction method according to any one of claims 1 to 6.

Images